Research On Monitoring Methods And Techniques Of Safety Driving Status Based On Kinect

Traffic safety problem is a major problem waiting to be solved in today's society.Human factor of driver is one of the main causes of traffic accidents.It has become the hot topic of intelligent transport systems that how to timely and effectively detect the unsafe driving status of driver to reduce traffic accidents caused by fatigue or distraction.As a new somatosensory sensor device,Kinect provides depth sensor,infrared camera,identifying human skeletal tracking and a series of new technologies.It promotes the practical application value in the monitoring of safety driving status for its low cost and good stability.In this thesis,Kinect is used to solve the problem of high cost and low efficiency of the current driving status monitoring.To reduce the occurrence of traffic accidents,the driving status of drivers is monitored comprehensively,and security warnings are provided timely to protect safety driving actively and effectively.In this thesis,the research of driving status monitoring is divided into two parts–the research of fatigue monitoring and the research of distracted behavior monitoring.To monitor driver fatigue effectively on different lighting environments,the thesis designs a plan of switching infrared images and color images according to the lighting conditions and takes appropriate pretreatment methods to improve the infrared image quality.On the basis of face recognition of Kinect,monitor the status of the head and mouth.The status of the eyes is monitored with integral projection method and PERCLOS.And then,use RBF neural network to fuse fatigue information of the head,mouth and eyes.Also,sliding windows method is used to improve fatigue monitoring results.At the same time,use the improved dispersed hidden Markov model(DHMM)to identify the dynamic behavior of the driver.Use joint vector angles to indicate driver behavior,and the angels are divided into two parts according to the left hand and the right hand.After discretization and compression,actions of the left hand and the right hand are recognised by DHMM,and together the results determine types of behaviors.And then,the experimental results show the effectiveness of the method of fatigue monitoring and behavior monitoring.On these bases,design and develop a simulation system of monitoring safe driving status based on Kinect: the system monitors the driver's safe driving status from multiple directions simultaneously;give scalable warnings according to the results of fatigue monitoring and behavior monitoring;at the same time,save the relevant data for further study.This thesis researches on methods and technologies of monitoring driving status for safety.Use multiple methods of comprehensive analysis and information fusion technology to improve the reliability and accuracy of monitoring.All in all,This thesis has carried on the beneficial exploration in the safeguard of safe driving.